Benchmarking 10GbE NAS-to-NAS Transfers: iperf, NFS, cp, and rsync
Comparing file-copy performance of a 15.17 GB file between Synology NAS systems over a 10GbE network on November 9-10, 2026, analyzing cp and rsync speeds.
Stock photo for illustration only, not from the actual event
- Benchmarked transfers of a 15.17 GB file across a Synology NAS 10GbE network
- iperf2 reached 9.42 Gbps, while cp achieved 479–494 MB/s and rsync hit 181 MB/s
- Tests were conducted on November 9-10, 2026, with SHA-256 verification
As one of the Synology NAS units neared capacity, a migration of data to another NAS over an existing 10GbE network was planned to establish realistic file-copy performance without altering NAS configurations. The tests took place between November 9 and 10, 2026, with all hostnames and data paths anonymized.
While iperf2 successfully reached 9.42 Gbps, file copy commands revealed a different performance landscape. The cp utility completed large-file copies at 479–494 MB/s, whereas a single rsync process achieved approximately 181 MB/s. Examining installed tools and system resource usage helped clarify these discrepancies.
Stock photo for illustration only, not from the actual event
The evaluation utilized an existing file sized at 16,285,199,445 bytes (15.17 GiB). The destination mounted the source share via NFSv3/TCP with an effective rsize of 128 KiB. An administrative PC coordinated the commands outside the direct data path, and all transferred data was verified against source bytes using SHA-256 checksums.
Reading the first 8 GiB of the source file to /dev/null completed in 13.43 seconds (640 MB/s) using dd. However, caching effects likely influenced this result since source RAID showed virtually no physical reads during later observation intervals, meaning 640 MB/s should not be interpreted as sustained source-HDD throughput.
Further contextual analysis reveals that a faster initial copy does not automatically equal an optimal migration workflow. While cp delivers higher raw speed for a single pass, rsync remains indispensable for production environments due to its robust support for incremental synchronization, interrupted transfers, and detailed file metadata preservation.
Testing concurrency by splitting the 8 GiB workload showed that running two dd processes increased aggregate throughput by about 20%, while two rsync processes boosted aggregate throughput by roughly 55%. This indicates that concurrency benefits multiple independent files rather than accelerating a single isolated file transfer.
Source: Dev.to
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment